shorter GRP3 units?
14 Jul 2014 21:44:27 UTC
Topic 197641
(moderation:
One of my hosts recently completed a CPU GRP3 result in about a third of the elapsed and CPU time expected from recent results. While I suspect it had experienced an error condition, the quorum partner did the work on an Nvidia GPU, also in about a third his recent time, and the WU pair validated and were awarded 177.69 credit, somewhat less than a third the recent standard value of 660.0.
Is this a sea change in Gamma Ray Pulsar #3 work?
shorter GRP3 units?
)
No, it's a change in the data file on which the latest tasks are based.
Until quite recently and for about 2 months, we've being doing tasks based on a data file named LATeah0092C.dat. In quick succession, we've now gone through 93C, 94C and are currently on 95C. Each time there is a new data file there will be tasks that Bernd refers to as "short ends" or "corner cases" which take a lot less than the standard time to crunch. If you search for those terms, I'm sure you'll find previous posts from Bernd where he has explained the situation.
There seem to be a bunch of "short ends" immediately after a new data file comes into use, which is why you'll be seeing them right now.
Cheers,
Gary.
RE: Each time there is a
)
Thanks. Searching for "short ends" got me to this Bernd message for example.